How to Start a Garden: A Frost-Date Guide With Cited Sources

Last verified: August 19, 2026. Frost data pulled from the NOAA NCEI 1991-2020 U.S. Climate Normals. Planting figures cited to named Cooperative Extension publications throughout.

Start a garden by working backward from one number: your last spring freeze date. Pick a spot with six to eight hours of direct sun, build or till a bed eight to twelve inches deep, then plant each crop by its published relationship to that freeze date, not by a calendar month. Hardy crops go in two to four weeks before it. Tender crops wait until two or more weeks after. Everything else on this page is detail, and I will walk you through it.

What does your USDA hardiness zone actually tell you?

Let’s start with the map everyone reaches for first.

Less than almost every gardening guide implies. The USDA Plant Hardiness Zone Map, updated November 15, 2023, is built from one variable: the average annual extreme minimum winter temperature, in 10-degree Fahrenheit zones split into 5-degree half zones. USDA states its purpose plainly on its own instruction page: “All Plant Hardiness Zone Maps (PHZM) should serve as general guides for growing perennial plants.”

That sentence is the whole point. The zone tells you whether a fig tree or a rosemary bush will survive your winter. It tells you nothing about when to direct-sow a carrot, because annual vegetables are planted, harvested, and finished inside a single growing season and never meet the extreme minimum the map measures.

USDA is also explicit about what the number excludes: “Wind, soil type, soil moisture, humidity, pollution, snow, and winter sunshine can greatly affect the survival of plants. Warm season heat and moisture balance are particularly important in this regard.” The agency further cautions that the map cannot capture the timing of cold, only its magnitude, and that microclimates smaller than the half-mile grid simply do not appear.

What changed in the 2023 map

Sources: USDA ARS press release, November 15, 2023 (last modified October 1, 2024); USDA ARS, Map Creation.

One more piece of USDA’s own language worth quoting, because a zone change gets treated as a mandate: “If your hardiness zone has changed in this edition of the USDA Plant Hardiness Zone Map (PHZM), it does not mean you should start removing plants from your garden or change what you are growing. What has thrived in your yard will most likely continue to thrive.”

When is your last frost date, and why is one date the wrong question?

This is where this guide diverges from the beginner articles I read while preparing it. NOAA’s National Centers for Environmental Information does not publish “the average last frost date.” It publishes a probability distribution: nine probability levels at six temperature thresholds (36, 32, 28, 24, 20 and 16 degrees Fahrenheit), inside the 1991-2020 U.S. Annual and Seasonal Climate Normals.

NC State Extension explains why 36 degrees is in the set at all: “A freeze can occur without a frost, and a frost can occur without a freeze. So there is no single temperature threshold for the occurrence of frost, but the risk of frost is relatively higher once temperatures fall below 36 degrees F.”

The three dates every gardener should know for their station

Read the spring numbers as exceedance probabilities:

• P90 is the aggressive date. In nine years out of ten you still get a freeze on or after it. Planting tender crops on it is a gamble.
• P50 is the median. This is what everyone means by “the average last frost date.” It fails half the time.
• P10 is the conservative date. Only one year in ten sees a freeze on or after it.

Last spring freeze at 32 degrees F, 1991-2020 normals

Pulled from the NOAA NCEI Annual/Seasonal Climate Normals (1991-2020) data service on August 19, 2026. Los Angeles Intl and Miami Intl publish no 32-degree spring value; their median freeze-free season is 365 days.

The spread is the story. Between the risk-averse date and the risk-seeking one there are 23 to 46 days. A single “average last frost date” throws that entire decision away. If you are setting out tomato transplants you cannot replace, use P10. If you are direct-sowing a packet of beans you are happy to re-sow, P50 or P90 is a reasonable bet.

Median freeze-free growing season

NOAA NCEI, 1991-2020 normals, median (P50) values. Compare this figure against the days-to-maturity column in the crop table below before you order seed.

Watch for the calendar wrap-around

In mild-winter stations the distribution crosses January 1 and the raw data produces dates that look absurd. Sea-Tac’s 28-degree P90 value is December 20 — meaning that in nine years out of ten the last hard freeze happened before New Year’s Day of that season. Phoenix’s 32-degree P90 is December 19 for the same reason. These are real values, correctly reported, that any site scraping the dataset without handling the wrap-around will publish as nonsense.

What is the difference between a frost date and a planting date?

Two independent gates, and Extension services treat them separately.

Gate one is air temperature, which the freeze probabilities above describe. It governs frost-tender crops.

Gate two is soil temperature, which governs germination regardless of what the air is doing. University of Maryland Extension: “Soil temperature is useful to know when direct-seeding beans, cucumbers, melons, and other crops that require soil temperatures of at least 60 degrees F to germinate. Cool, wet soils slow germination and make seeds more susceptible to rot.”

Oklahoma State’s two-line rule is the simplest version: cool-season vegetables want at least 40 degrees F at seed depth; warm-season vegetables want at least 50 degrees F. But note that University of Arizona Extension quotes Ohio State recommending warm-season crops go in only after soil reaches 55 to 60 degrees F — a 5 to 10 degree higher bar for the same crop class from two land-grant institutions.

Colorado State adds a measurement protocol most gardeners never hear: read soil temperature at 4 inches deep, at 8 a.m. Beans are the exception and get measured at 6 inches.

Arizona Extension also makes an observation worth more than most planting advice: “You may find that waiting and planting after June 1 allows the plants to grow/germinate rapidly and they will often outgrow the plants that were planted earlier using a season extender.”

When do you plant each vegetable?

Now, the part you came for.

The table below is built entirely from Cooperative Extension publications and expresses timing relative to your last spring freeze date rather than a calendar month, so it works in every zone. Where extensions disagree, I print both figures and name the disagreement in the final column. I do not average them, because the average is a number no institution published.

One caveat that explains many of the conflicts, from University of Maryland Extension (May 2026): for tomatoes, peppers, eggplants, muskmelons, watermelons, broccoli, cabbage, cauliflower and Brussels sprouts, days-to-maturity usually counts from transplanting. For every other annual crop it counts from direct seeding. Two sources quoting different day counts are often measuring from different starting points.

Hardy — plant 2 to 4 weeks before your last freeze

Colorado State: hardy vegetables “grow with daytime temperatures as low as 40 degrees F and may survive a frosty nip.”

Semi-hardy — plant 0 to 2 weeks before your last freeze

Colorado State: these “grow with minimum daytime temperatures of 40 to 50 degrees F but are less tolerant of a frosty night.”

Tender — plant around your last freeze date

Colorado State: tender vegetables “grow with a daytime temperature above 55 degrees F and are intolerant of frost.”

Very tender — plant 2 or more weeks after your last freeze

Colorado State: these “need daytime temperatures above 60 degrees F, and prefer temperatures of 70 to 95 degrees F. A week of daytime temperatures below 55 degrees F may stunt the crop.”

Timing column: UNH Extension, When to Plant Your Vegetable Garden (no date shown on page). Frost class, soil temperatures and CSU day counts: Colorado State CMG GardenNotes 720, Vegetable Planting Guide, reviewed September 2024. Harrington/UC Davis germination temperatures via Oregon State Extension, published April 2013, reviewed 2024. USU day ranges: Utah State Extension Planting Guide (no date shown). Garlic: UMN Extension, Growing garlic, reviewed 2024. Basis note: UMD Extension, When to Plant Vegetables, May 2026. Tier definitions also compared against Oklahoma State HLA-6004, February 2021 and University of Illinois, March 26, 2018. Download this table as CSV.

Does companion planting actually work?

Some of it. Most of the charts do not.

Washington State University Extension horticulturist Linda Chalker-Scott is the bluntest institutional voice on this: “There is no scientific basis, but for any of the several lists that exist describing ‘traditional companion plants’. Like horoscopes, these lists may be fun to use, but they should not be perceived or promoted as scientifically valid any more than astrology.” Her stated bottom line is that traditional companion plant charts have “entertainment, not scientific, value.”

University of Arizona Extension puts it more gently and explains why the literature is thin: “Companion planting is not a proven science. The basis for this type of gardening comes from folklore and experimentation.” It names the two obstacles as control of variables and repeatability. UMass Extension warns that “choosing a companion planting scheme not grounded in science may result in disappointment.”

The most useful institutional source is University of Minnesota Extension, which is the only extension companion-planting page I found carrying a full literature reference list. Its framing: “While gardening charts online will provide you with long lists of plants that repel insects, these are not always accurate or backed by research.”

Companion planting, graded by evidence

Primary sources: UMN Extension, Companion planting in home gardens (reviewed 2026); UF/IFAS ENY-056, Marigolds for Nematode Management (revised April 2016, reviewed January 2024); WSU, The Myth of Companion Plantings (Chalker-Scott); University of Arizona, The Art of Companion Planting, April 2023; UMass Extension, last updated July 7, 2012.

Raised bed or in-ground?

Utah State Extension’s peer-reviewed guidance: a raised bed box should be at least 6 to 12 inches high, and beds shallower than 12 inches “should be constructed with no bottom to allow plant roots access to existing soil below the box.” Till 6 to 12 inches beneath the box before filling. Keep beds 3 to 4 feet wide so you can reach the middle from either side. Colorado State suggests 2 to 4 foot beds with 2-foot walkways.

For soil, Utah State gives two routes: amend native topsoil with 30 to 50 percent well-composted organic matter by volume, or blend equal parts composted organic matter, a drainage material (coarse sand, vermiculite, pumice or perlite) and native soil. On bagged soilless mixes it warns they are expensive and “may become nutrient deficient over time.”

If your organic matter is not fully composted, Utah State’s nitrogen correction is specific: for every inch of added woody material, incorporate a quarter pound of actual nitrogen per 100 square feet.

Two trade-offs Utah State names that most raised-bed content omits: raised beds warm faster in spring, which is the selling point, but “higher soil temperatures will increase evaporative moisture loss, requiring more frequent irrigation during the hot summer months.” And flatly: “A raised bed garden must be irrigated.”

Is treated lumber safe for a food bed?

Start with the regulatory record, because almost every garden article states it wrong. CCA was never banned by EPA. Per EPA’s chromated arsenicals page, last updated December 22, 2025: “In December 2003, chromated arsenicals manufacturers voluntarily discontinued manufacturing chromated arsenicals-treated wood products for homeowner uses.” It was a voluntary cancellation, for residential uses only, and CCA remains registered for commercial applications today. EPA also states: “EPA did not find health risks of concern for the general public.” Its risk finding of concern is for workers in wood treatment facilities.

The date gets published three different ways depending on what is being described: EPA announced the phase-out in 2002, manufacture for homeowner uses ended in December 2003, and retail availability ended around 2004. All three are defensible. Say which you mean.

Modern residential treated lumber is a different chemistry. Purdue Extension describes alkaline copper quaternary (ACQ) as “higher in copper than CCA but free of arsenic,” and copper azole as an above-ground-only treatment. University of Maryland Extension, updated March 29, 2024: “Micronized Copper Azole (MCA) is currently the most widely available type of wood preservative for residential use. It does not contain arsenic, but it does contain copper.”

The newest and most direct research is Oregon State Extension publication EM 9879, revised April 2026, which measured copper in soil and in the vegetables themselves. Its findings: copper increases were small and limited to soil within one inch of the wood, and there was “no increase in copper concentration in each of the vegetables and herbs grown in raised beds made with pressure-treated lumber in any of the growing seasons.” Its conclusion, with the geographic scope OSU itself wrote into it: “Willamette Valley gardeners should not be concerned with copper accumulation in vegetables and herbs grown in raised beds constructed with copper azole-treated wood intended for ‘ground contact’ use.”

Three extensions, three answers on plastic barriers

Utah State’s page predates the Oregon State field trial by 14 years and states a leaching mechanism OSU’s measurements do not support at the plant-uptake level. I report all three with their dates rather than resolving it silently. Avoid the old chemistries regardless: UMD lists CCA, creosote and penta-treated lumber as materials to avoid, and notes that concrete block often contains fly ash, “a byproduct of burning coal” that “contains heavy metals.”

On tires as bed material, I found no primary source in either direction and make no claim.

What should a first-year garden actually grow?

Work from your freeze-free season length in the table above and the days-to-maturity column, then bias toward crops with wide germination windows and short maturity. Radish (30 days), lettuce (60), spinach (40), kohlrabi (50) and turnip (50) all mature fast and tolerate cold soil, which means a mistake costs you weeks rather than a season. Peas go in the moment the ground thaws.

The crops that punish beginners are the ones with long transplant lead times and narrow temperature windows: celery needs 10 to 12 weeks indoors before it goes out, onions from seed need 8 to 10, and peppers need 8 to 10 while also demanding night temperatures above 60 to 65 degrees F once outside.

Six to eight hours of direct sun is the siting figure Utah State publishes for maximum fruit and foliage production; UMD sets the floor at six hours. Below that you are choosing leaf crops, not fruiting ones.

Frequently Asked Questions

Does my USDA hardiness zone tell me when to plant vegetables?

No. USDA scopes the hardiness zone map to perennials in its own words — it is a winter-survival map, not a planting calendar. Use your last frost date to time vegetable planting instead.

Is there a single “last frost date” for my area?

No. NOAA publishes nine probability levels at six temperature thresholds, and the spread between the cautious date and the aggressive one runs 23 to 46 days depending on where you live.

How deep should a garden bed be?

Eight to twelve inches, tilled or built up, in a spot with six to eight hours of direct sun.

Are companion planting charts based on real science?

Mostly not. Washington State University Extension calls the traditional companion planting charts entertainment rather than science. A small set of pairings does have research behind them.

Final Thoughts

Remember, your zone tells you what survives winter, and your station’s freeze dates tell you when to plant. Get those two things from the right sources, plant by the four hardiness groups, and let the companion-planting folklore go. That is a first garden that feeds you.

A note on sources

Every number on this page comes from a named primary source with its publication or review date: USDA ARS for hardiness zones, NOAA NCEI’s 1991-2020 Climate Normals data service for freeze probabilities, EPA for the treated-lumber regulatory record, and Cooperative Extension publications from Colorado State, UNH, Utah State, Oklahoma State, Illinois, Maryland, Minnesota, Massachusetts, Arizona, Oregon State, Purdue, Washington State and UF/IFAS for crop and companion-planting guidance. Where two institutions disagree, both figures appear with attribution. Where I could not source a claim, it is absent or labeled. Researched and human-edited; no field trials were conducted for this article.

A note on what is missing. I deliberately left out several widely published claims because no institutional source supports them: the pea-and-onion warning, the carrot-and-dill warning, the marigold-and-onion-maggot claim, and the assertion that Three Sisters out-yields monoculture. I also do not publish a juglone (black walnut) section, because I have not yet read the Penn State publication others attribute it to. These will be added when they can be cited.